JP2002131019A - Measuring method of micro dimension - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten a measuring time with calibrating a dimension measured value responding to two-dimensional distortion in a measured screen, without fluctuations of the value owing to a measuring position in the screen and without slight adjustment by a XY stage mechanism. SOLUTION: In a measuring instrument of micro dimension to image a measured object by using an optical microscope and a two-dimensional sensor, to extract signal positions of the two points matching to a predetermined brightness level from the obtained video signal and to calculate and measure dimensions of the object based on the position difference information between the two points, a measured position difference information Nab is multiplied by a factor k determined from a measuring scaling factor of the microscope and a distance to the object, when acquiring a dimension W of the responded measured object, the measured dimension W is acquired with linearity calibration by setting W=k×D(Xn, Ym)/rPy responding to a two-dimensional distortion ratio rPy at the measured screen position D (Xn, Ym).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for calibrating the linearity of an apparatus for measuring the size of an object without contact using a two-dimensional sensor such as a TV camera.

[0002]

2. Description of the Related Art The configuration of a conventional basic dimension measuring device (for example, Japanese Patent Publication No. 6-103168) is shown in FIG.
As shown in (1), a spatial image of a subject (object to be measured) 2 projected by an optical microscope 1 is captured by a TV camera 3, and a dimension of a desired portion is measured by a dimension measurement arithmetic processing unit 4 ′. Some display the image of the device under test 2 and the dimension measurement values. Here, as shown in FIG. 3A, the luminance distribution on one horizontal scanning line Li in the monitor image 5 'of the device under test 2 imaged by the TV camera 3 is obtained by dividing the video signal corresponding to the scanning line Li by N decomposition. FIG. 3 shows the relationship between each pixel position and each luminance.
The luminance-pixel characteristics of (b) are obtained. As a conventional processing method, the dimension can be measured from this characteristic as follows. In FIG. 3, the maximum luminance level 5 in the luminance distribution
1 is set to 100%, the minimum luminance level 52 is set to 0%, and 5
The position difference Nab between the a-th pixel and the b-th pixel corresponding to the 0% luminance level 53 is obtained, and this position difference Nab is calculated.
Is multiplied by a coefficient k determined by the measurement magnification of the microscope 1 at this time and the subject distance from the TV camera 3 to the device under test 2.
The corresponding dimension value W of the device under test 2 is determined. W = k ×
Nab

[0003]

As described above, in the prior art, the coefficient k is the same no matter where the measurement is made on the screen, and the distortion of the optical system and the distortion of the sensor are not taken into account. The measured value fluctuates by about 0.5% depending on the measurement location.
For example, when the DUT 2 shown in FIG. 3A is measured at nine locations on the screen as shown in FIG. 3C, the dimension values fluctuate as shown in Table 1 below.

[0004]

[Table 1] For this reason, XY measurement is performed so that measurement is performed at the same location in the screen.
The measurement is performed after the object to be measured 2 is finely adjusted on the stage, and the object to be measured 2 needs to be finely adjusted by a high-precision XY stage mechanism, which complicates the configuration and requires a long measurement time. The present invention eliminates these drawbacks, calibrates the dimension measurement values according to the two-dimensional distortion in the measurement screen, eliminates fluctuations in the measurement values due to measurement points in the screen, and does not require fine adjustment by the XY stage mechanism. It is intended to shorten the measurement time.

[0005]

According to the present invention, in order to achieve the above object, an object to be measured is imaged using an optical microscope and a two-dimensional sensor, and a predetermined luminance level is obtained from an obtained video signal. In a minute dimension measuring apparatus that extracts two signal positions and calculates and measures the dimensions of the object to be measured based on the positional difference information between the two points, the measured positional difference information Nab includes:
At this time, when the measurement magnification of the optical microscope and the coefficient k determined by the subject distance are multiplied to obtain the corresponding dimension W of the DUT, the two-dimensional distortion ratio rPy at the measured screen position D (Xn, Ym) The above measured dimension W is expressed as W = k
The linearity is calibrated as × D (Xn, Ym) / rPy. In this way, the method of measuring the two-dimensional distortion in the measurement screen and calibrating the measured value in accordance with the two-dimensional distortion eliminates the disadvantage that the measured value fluctuates by about 0.5% depending on the measurement location in the screen. Can be solved.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The construction and operation of an embodiment of the present invention will be described below in detail with reference to FIGS. 1 (a) to 1 (c). Here, for example, when the visual field of the measurement screen is X = 20
μm, Y = 20 μm, the number of pixels of the sensor is 10
In an apparatus of 00 × 1000, the above coefficient k is k = 20
/1000=0.02 μm. In the conventional apparatus, as described above, the measured line width is set to Nab = D (X, Y), and the dimension W is set as follows: W = k × Nab = k × D (X, Y)
Seeking in. In the present invention, the measured line width is defined as Nab =
D (Xn, Ym), and in addition to this coefficient k, using the screen distortion ratio rPy at the measured screen position (Xn, Ym), the dimension W is expressed as W = k × D (Xn, Ym) / rPy. Ask.

Hereinafter, a method for obtaining the screen distortion ratio rPy of the present invention will be described in detail. First, a screen distortion registration method will be described with reference to FIG. An aerial image of the object (object to be measured) 2 projected by the optical microscope 1 is captured by a TV camera 3, and a dimension of a desired portion is electrically measured by a dimension measurement arithmetic processing unit 4. The second image and the dimension measurements are displayed. The movement of the device under test 2 is performed by the XY electric stage (not shown) from the dimension measurement arithmetic processing device 4.
And instructed. Here, the XY electric stage is 0.1
It moves at a resolution of μm. Also, the DUT 2
Is a reference sample of a known size W. Also, as shown in FIG. 1B, n locations in the X direction and m locations in the Y direction in the measurement screen are defined as measurement locations (Xn, Ym). Here, first, the XY electric stage is moved to the measurement location (X1, Y1), and the measured value D is measured.
(X1, Y1) is obtained. Next, the XY electric stage is moved to the measurement location (X1, Y2), and the measurement value D (X1, Y2) is obtained. Further, the XY motorized stage is sequentially moved to the measurement location (X
1, Yn) to obtain a measured value D (X1, Yn).
Next, the XY electric stage is moved to the measurement location (X2, Y1), and the measurement value D (X2, Y1) is obtained. Then, the XY electric stage is moved to the measurement point (X2, Y2), and the measured value D
(X2, Y2) is obtained. Further, the XY electric stage is sequentially moved to the measurement location (X2, Yn), and the measured value D (X
2, Yn). Finally, the center of the reference sample is moved to the measurement location (Xn, Ym) by the XY motorized stage, and the measured value D (Xn, Ym) is obtained.

[0008] Here, a coefficient k for setting the measured value D (X1, Y1) of the measurement point (X1, Y1) measured as described above to a known dimension W is obtained from the following equation. k = W / D (X
1, Y1) That is, this k is the measurement location (X1, Y1)
Is a coefficient with W being the measured value D (X1, Y1) of
W = k × D (X1, Y1) Then, the measured value D (Xn, Ym) is converted to the measured value D (X1, Y1).
1) to divide the ratio r (X1, Y1) from the ratio r (X1, Y1).
(Yn, Ym) is obtained. For example, when measurement is performed at n = 3 and m = 3 locations, n = 1, 2, 2 on the Y coordinate of m = 1
3 at the X coordinate, r (X1, Y1), r (X2, Y1),
r (X3, Y1) is obtained. That is, the coordinates of X1, X2, and X3 and the ratios r (x1, Y1), r (X2, Y1), and r (X
From (3, Y1), the quadratic equation r (Y1) = A1 · X ² +
B1 · X + C1 coefficients A1, B1, C1 calculated by the calculation, a quadratic equation r (Y1) = A1 · X 2 + B1 · X + C
Register 1. Next, on the Y coordinate of m = 2, n = 1,
The ratio r (X1, Y2) at the X coordinate of 2, 3 and r (X2, Y
2), r (X3, Y2) is obtained.

Then, similarly, a quadratic equation r (Y2)
= To register the ^{A2 · X 2 + B2 · X} + C2. Also, m
= R ratio on the X coordinate of n = 1,2,3 on the Y coordinate of = 3
(X1, Y3), r (X2, Y3) and r (X3, Y3) are obtained. Then, similarly, the quadratic equation r (Y3) = A3 ·
Register X ² + B3 · X + C3. As above,
r (Y1), r (Y2), r (Y) of the Y coordinate of m = 1, 2, 3
3) Register. That is, when the measurement is performed at n and m points, the following equation (n-1) can be formed at the maximum.

[0010] m = 1,2,2 registered as above
The quadratic equation r (Y1), (Y2), r (Y3) of the Y coordinate of 3
Is expressed by including the Y coordinate, the registered quadratic equation r
The relationship between (Ym) and the measurement point P (Px, Py) is shown in FIG.
become that way. That is, the measurement point P (Px, Py) when measured at, m = 1 in the equation r (Y1) = A1 · X 2 + B1 · X + C
1 Equation of m = 2 r (Y2) = A2 · X ² + B2 · X + C
2 Equation of m = 3 r (Y3) = A3 · X ² + B3 · X + C
3, by substituting X = Px, obtain r (Y1) = A1 · Px 2 + B1 · Px + C1 r (Y2) = A2 · Px 2 + B2 · Px + C2 r (Y3) = A3 · Px 2 + B3 · Px + C3. Here, r (Y1), r (Y2) and r (Y3) are functions of the Y coordinate, and a quadratic equation rPy = Ap · Y ²
By substituting Y = Y1, Y2, and Y3 into + Bp.Y + Cp, Ap, Bp, and Cp are obtained. Then, the finally determined ratio rPy is represented by Y = Py, rPy = Ap · Y
^By substituting ² + Bp · Y + Cp, rPy = Ap ·
Yp ² + Bp · Yp + Cp is obtained. Accordingly, the line width W to be obtained is obtained by dividing k × D (X1, Y1) by the ratio rPy, and W = k × D (X1, Y1) / rPy
Thus, the line width W with the screen distortion corrected can be obtained.

The above is verified using actual numerical values. Using a two-dimensional sensor with 1000 × 1000 pixels,
A line width of 1.05 μm is measured in a screen visual field of 20 μm × 20 μm, and it is assumed that the pixels D (Xn, Ym) of each line width are as shown in Table 1 described above. Here, a coefficient k for setting the measured value D (X1, Y1) of the measurement point (X1, Y1) to the known dimension W = 1.050 is obtained. k = W / D (X1, Y1) = 1.050 / 50.52 = 0.020784 That is, as described below, this k is the measured value D (X1, Y1) of the measurement point (X1, Y1). , And a coefficient for setting the line width dimension W. W = k × D (X1, Y1) = 0.020784 × 50.52 = 1.050 Next, as shown in Table 2 below, the measured value D (Xn, Xm) is converted to the measured value D (X1, Y1 ), And the ratio r (X1, Y
The ratio r (Yn, Ym) is obtained from 1).

[0012]

[Table 2] Here, on the Y coordinate of m = 1, the ratios r (X1, Y1), r (X2, Y1), r (X
3, Y1).

Next, the coordinates of X1, X2 and X3 and the ratio r
(X1, Y1), r (X2, Y1), r (X3, Y1)
From the quadratic equation r (Y1) = A1 · X ² + B1 · X +
The coefficients A1, B1, and C1 of C1 are obtained by the following calculation. 1.0000 = A1 · 200 ² + B1 · 200 + C1 1.0022 = A1 · 550 ² + B1 · 550 + C1 1.0018 = A1 · 800 ² + B1 · 800 + C1 1.0000 = 40000A1 + 200B1 + C1 ··· 302500A1 + 550B1 + C1 (2) 1.0018 = 640000A1 + 800B1 + C1 (3) From C1 = 1.0000- (40000A1 + 200B1), Expressions (2) and (3) are as follows: 1.0022 = 302500A1 + 550B1 + 1.0000 -(40000A1 + 200B1) 0.0022 = 262500A1 + 350B1 ... (4) 1.0018 = 640000A1 + 800B1 + 1.0000-(40000A1 + 200B1) 0.0018 = 600000A1 + 600B1 ... (5) B1 = (0.0018- 600000A1) / 600 is calculated by the formula (4) ), 0.0022 = 262500A1 + 350 (0.0018-600000A1) / 600 0.0022-0.0018 (350/600) = (262500-600000 (350/600)) A1 0.0022-0.00105 = (262500−350,000) A1 A1 = −0.00115 / 87500 (6) B1 = (0.0018−600000A1) / 600 is substituted for Equation (6), and B1 = ( 0.0018 + 0.0078857) / 600 B1 = 0.0096857 / 600 (7) C1 = 1.0000− (40000A1 + 200B1), and substituting the expressions (6) and (7) into C1 = 1.0000-(40000 (-0.00115 / 87500) + 200 • 0.0096857 / 600) = 1.0000-(-0.0005257 + 0.00328568) C1 = 0.999729 .. (8) As described above, the coefficient A of the above equations (6), (7) and (8)
1, B1 and C1 are calculated, and a quadratic equation of r (Y1) is registered. r (100) = − 0.00115 / 87500X ² + (0.0096857 /
600) X + 0.999729

Similarly, on the Y coordinate of m = 2,
Ratio r (X1, Y2), r of X coordinate locations where n = 1, 2, 3
(X2, Y2) and r (X3, Y2) are obtained. 0.9899 = 40000A2 + 200B2 + C2 0.9901 = 302500A2 + 550B2 + C2 0.9912 = 640000A2 + 800B2 + C2 and, quadratic equation r (Y2) = A2 · X 2 + B2 · X
The coefficients A2, B2, and C2 of + C2 are obtained in the same manner as r (Y1), and a quadratic equation of r (Y2) is registered. r (500) = − 0.0005583 / 87500 · X ² − (0.00253
/600)X+0.999948 Similarly, on the Y coordinate of m = 3, n = 1, 2, 2,
3, the ratio r (X1, Y3), r (X2, Y
3), r (X3, Y3) is obtained. 0.9956 = 40000A3 + 200B3 + C3 0.9966 = 302500A3 + 550B3 + C3 0.9976 = 640000A3 + 800B3 + C3 and, quadratic equation r (Y3) = A3 · X 2 + B3 · X
The coefficients A3, B3, and C3 of + C3 are obtained in the same manner as r (Y1), and a quadratic equation of r (Y3) is registered. r (900) = 0.000167 / 87500 · X 2 -0.000857 / 6
00 · X + 0.995238 When the quadratic equations r (Y1), (Y2), and r (Y3) of the Y coordinate of m = 1, 2, 3 registered as described above are expressed by including the Y coordinate, The relationship between the registered quadratic equation r (Ym) and the measurement point P (Px, Py) is as shown in FIG.

As an example, the measurement point P (Px, Py) =
When measured at the point P (300,400), the equation of m = 1 r (100) = (− 0.00115 / 87500)
X ² + (0.0096857 / 600) X + 0.999729 Equation of m = 2 r (500) = (− 0.00005583 / 87500)
X ² − (0.00253 / 600) X + 0.999948 m = 3 equation r (900) = (0.00000167 / 87500) X
² - (0.000857 / 600) X + 0.995238 , obtain X = 300 substitutes r (100) = 1.00958 r ( 500) = 0.98980 r (900) = 0.99583. r (Yl), r (Y2 ), r (Y3) is a function of the Y coordinate, the quadratic equation ^{rPy = Ap · Y 2 + Bp} · Y + Cp
Is substituted for Y = Y1, Y2, Y3, and 1.00958 = Ap · 100 ² + Bp · 100 + Cp 0.99880 = Ap · 500 ² + Bp · 500 + Cp 0.99583 = Ap · 900 ² + Bp · 900 + Cp Bp and Cp are obtained. Ap = 8.04063E- ⁸ , Bp = -9.75E- ⁵ , Cp = 1.01
8445 The finally determined ratio rPy is expressed as follows: Y = Py = 400
Substituting rPy = Ap · Y ² + Bp · Y + Cp, Is determined by W = 1.05 μm, k = 0.020784, X1 = Px
= 800, Y2 = Py = 400, W = k × D (300,400) / rPy 1.05 = 0.020784 × D (300,400) /0.99231 D (300,700) = (1 .05 / 0.020784) x 0.999231 = 50.1
It can be expected to be 3 pixels, W = k × D (300,400)
By using rPy = 0.99231 in the formula, W
= 0.020784 × 50.13 / 0.92311 = 1.05 μm.

In the above description, an example of a quadratic equation of three-point measurement has been shown. (1) Four or more points are measured, and a quadratic equation is obtained by the least square method. (2) Four points are measured, and a cubic equation is obtained by the least square method. (3) Five or more points are measured, and a cubic equation is obtained by the least square method. (4) The n points are measured, and the (n-1) -th order equation is obtained by the least square method. (5) Measure at least (n + 1) points, and use the least squares method to calculate (n−
1) Find the following equation. This is also according to the present invention.

[0017]

(1) At the time of registration, a polynomial equation is created from the measured points and the linearity at an arbitrary measurement point can be calibrated, so that the same dimensions can be obtained regardless of the measurement anywhere on the screen. (2) The linearity calibration can be automatically registered simply by setting a known reference sample on the XY motorized stage, and the reliability is independent of the operation and judgment of the operator. (3) Since it is automatically read out and calculated automatically at an arbitrary measurement point in the screen, it is simple and reliable without depending on the operation and judgment of the operator.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the configuration and operation of a minute dimension measuring apparatus according to the present invention.

FIG. 2 is a diagram showing a configuration of a conventional minute dimension measuring device.

FIG. 3 is a diagram illustrating a conventional measurement method.

[Explanation of symbols]

1: optical microscope, 2: measured object, 3: TV camera, 4:
Dimension measurement arithmetic processing unit, 5: TV monitor, 41: frame memory, 42: high resolution memory, 43: CPU

Claims (2)

[Claims] 1. An object to be measured is imaged using an optical microscope and a two-dimensional sensor, and two signal positions corresponding to a predetermined luminance level are extracted from an obtained video signal, and a positional difference between the two points is extracted. In the minute dimension measuring device that calculates and measures the dimension of the object to be measured based on the information, the measured position difference information Nab includes:
At this time, when the measurement magnification of the optical microscope and the coefficient k determined by the subject distance are multiplied to obtain the corresponding dimension W of the DUT, the two-dimensional distortion ratio rPy at the measured screen position D (Xn, Ym) A small dimension measuring method, wherein the measured dimension W is obtained by linearity calibration. 2. The method according to claim 1, wherein the measured dimension W is calculated according to the two-dimensional distortion ratio rPy by W = k × D (X
(n, Ym) / rPy.